New system can recreate faces from brain signals

A new artificial intelligence machine that can identify criminals from a victim's thoughts may be closer to reality, say scientists who have successfully recreated images of a human face by analysing brain activity of monkeys.When you look at photos of faces, your brain is able to instantly identify the ones that you know - whether it is your mother or favourite celebrity - and distinguish among those that you have never seen before. 

In recent years, neuroscientists have begun to peek inside the brain's black box to understand how the brain is able to recognise and perceive faces. Researchers have now cracked the code for facial identity in the primate brain. "We've discovered that this code is extremely simple," said Doris Tsao, a professor at the California Institute of Technology in the US.

"A practical consequence of our findings is that we can now reconstruct a face that a monkey is seeing by monitoring the electrical activity of only 205 neurons in the monkey's brain," said Tsao. "One can imagine applications in forensics where one could reconstruct the face of a criminal by analysing a witness's brain activity," he said. Earlier research used fMRI in humans and other primates to identify six areas of the brain that are responsible for identifying faces.

These areas, called face patches, are located in the inferior temporal (IT) cortex. Further studies showed that these areas are packed with specific nerve cells that fire action potentials much more strongly when seeing faces than when seeing other objects. They called these neurons face cells. 

Previously, some experts in the field believed that each face cell in the brain represents a specific face, but this presented a paradox, said Tsao. "You could potentially recognise 6 billion people, but you don't have 6 billion face cells in the IT cortex. There had to be some other solution," he said. In the current study, published in the journal Cell, researchers found that rather than representing a specific identity, each face cell represents a specific axis within a multidimensional space, which they call the face space.

In the same way that red, blue, and green light combine in different ways to create every possible colour on the spectrum, these axes can combine in different ways to create every possible face. Using macaque monkeys as a model system, the researchers inserted electrodes into the brains that could record individual signals from single face cells within the face patches. They then developed an algorithm that could decode additional faces from neural responses.

They could now show the monkey an arbitrary new face, and recreate the face that the monkey was seeing from electrical activity of face cells in the animal's brain. When placed side by side, the photos that the monkeys were shown and the faces that were recreated using the algorithm were nearly identical. Face cells from only two of the face patches were enough to reconstruct the faces.